nokogiri 1.11.7 → 1.12.0.rc1

data/ext/nokogiri/{html_document.c → html4_document.c}

@@ -1,6 +1,6 @@
 #include <nokogiri.h>
 
-VALUE cNokogiriHtmlDocument ;
+VALUE cNokogiriHtml4Document ;
 
 static ID id_encoding_found;
 static ID id_to_s;
@@ -34,7 +34,7 @@ rb_html_document_s_new(int argc, VALUE *argv, VALUE klass)
  *  read_io(io, url, encoding, options)
  *
  * Read the HTML document from +io+ with given +url+, +encoding+,
- * and +options+.  See Nokogiri::HTML.parse
+ * and +options+.  See Nokogiri::HTML4.parse
  */
 static VALUE
 rb_html_document_s_read_io(VALUE klass, VALUE rb_io, VALUE rb_url, VALUE rb_encoding, VALUE rb_options)
@@ -92,7 +92,7 @@ rb_html_document_s_read_io(VALUE klass, VALUE rb_io, VALUE rb_url, VALUE rb_enco
  *  read_memory(string, url, encoding, options)
  *
  * Read the HTML document contained in +string+ with given +url+, +encoding+,
- * and +options+.  See Nokogiri::HTML.parse
+ * and +options+.  See Nokogiri::HTML4.parse
  */
 static VALUE
 rb_html_document_s_read_memory(VALUE klass, VALUE rb_html, VALUE rb_url, VALUE rb_encoding, VALUE rb_options)
@@ -153,13 +153,13 @@ void
 noko_init_html_document()
 {
   assert(cNokogiriXmlDocument);
-  cNokogiriHtmlDocument = rb_define_class_under(mNokogiriHtml, "Document", cNokogiriXmlDocument);
+  cNokogiriHtml4Document = rb_define_class_under(mNokogiriHtml4, "Document", cNokogiriXmlDocument);
 
-  rb_define_singleton_method(cNokogiriHtmlDocument, "read_memory", rb_html_document_s_read_memory, 4);
-  rb_define_singleton_method(cNokogiriHtmlDocument, "read_io", rb_html_document_s_read_io, 4);
-  rb_define_singleton_method(cNokogiriHtmlDocument, "new", rb_html_document_s_new, -1);
+  rb_define_singleton_method(cNokogiriHtml4Document, "read_memory", rb_html_document_s_read_memory, 4);
+  rb_define_singleton_method(cNokogiriHtml4Document, "read_io", rb_html_document_s_read_io, 4);
+  rb_define_singleton_method(cNokogiriHtml4Document, "new", rb_html_document_s_new, -1);
 
-  rb_define_method(cNokogiriHtmlDocument, "type", rb_html_document_type, 0);
+  rb_define_method(cNokogiriHtml4Document, "type", rb_html_document_type, 0);
 
   id_encoding_found = rb_intern("encoding_found");
   id_to_s = rb_intern("to_s");

data/ext/nokogiri/{html_element_description.c → html4_element_description.c}

@@ -1,6 +1,6 @@
 #include <nokogiri.h>
 
-VALUE cNokogiriHtmlElementDescription ;
+VALUE cNokogiriHtml4ElementDescription ;
 
 /*
  * call-seq:
@@ -272,21 +272,23 @@ get_description(VALUE klass, VALUE tag_name)
 void
 noko_init_html_element_description()
 {
-  cNokogiriHtmlElementDescription = rb_define_class_under(mNokogiriHtml, "ElementDescription", rb_cObject);
-
-  rb_define_singleton_method(cNokogiriHtmlElementDescription, "[]", get_description, 1);
-
-  rb_define_method(cNokogiriHtmlElementDescription, "name", name, 0);
-  rb_define_method(cNokogiriHtmlElementDescription, "implied_start_tag?", implied_start_tag_eh, 0);
-  rb_define_method(cNokogiriHtmlElementDescription, "implied_end_tag?", implied_end_tag_eh, 0);
-  rb_define_method(cNokogiriHtmlElementDescription, "save_end_tag?", save_end_tag_eh, 0);
-  rb_define_method(cNokogiriHtmlElementDescription, "empty?", empty_eh, 0);
-  rb_define_method(cNokogiriHtmlElementDescription, "deprecated?", deprecated_eh, 0);
-  rb_define_method(cNokogiriHtmlElementDescription, "inline?", inline_eh, 0);
-  rb_define_method(cNokogiriHtmlElementDescription, "description", description, 0);
-  rb_define_method(cNokogiriHtmlElementDescription, "sub_elements", sub_elements, 0);
-  rb_define_method(cNokogiriHtmlElementDescription, "default_sub_element", default_sub_element, 0);
-  rb_define_method(cNokogiriHtmlElementDescription, "optional_attributes", optional_attributes, 0);
-  rb_define_method(cNokogiriHtmlElementDescription, "deprecated_attributes", deprecated_attributes, 0);
-  rb_define_method(cNokogiriHtmlElementDescription, "required_attributes", required_attributes, 0);
+  cNokogiriHtml4ElementDescription = rb_define_class_under(mNokogiriHtml4, "ElementDescription", rb_cObject);
+
+  rb_undef_alloc_func(cNokogiriHtml4ElementDescription);
+
+  rb_define_singleton_method(cNokogiriHtml4ElementDescription, "[]", get_description, 1);
+
+  rb_define_method(cNokogiriHtml4ElementDescription, "name", name, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "implied_start_tag?", implied_start_tag_eh, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "implied_end_tag?", implied_end_tag_eh, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "save_end_tag?", save_end_tag_eh, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "empty?", empty_eh, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "deprecated?", deprecated_eh, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "inline?", inline_eh, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "description", description, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "sub_elements", sub_elements, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "default_sub_element", default_sub_element, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "optional_attributes", optional_attributes, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "deprecated_attributes", deprecated_attributes, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "required_attributes", required_attributes, 0);
 }

data/ext/nokogiri/{html_entity_lookup.c → html4_entity_lookup.c}

@@ -1,17 +1,17 @@
 #include <nokogiri.h>
 
-static VALUE cNokogiriHtmlEntityLookup;
+static VALUE cNokogiriHtml4EntityLookup;
 
 /*
  * call-seq:
  *  get(key)
  *
- * Get the HTML::EntityDescription for +key+
+ * Get the HTML4::EntityDescription for +key+
  */
 static VALUE
 get(VALUE _, VALUE rb_entity_name)
 {
-  VALUE cNokogiriHtmlEntityDescription;
+  VALUE cNokogiriHtml4EntityDescription;
   const htmlEntityDesc *c_entity_desc;
   VALUE rb_constructor_args[3];
 
@@ -24,14 +24,14 @@ get(VALUE _, VALUE rb_entity_name)
   rb_constructor_args[1] = NOKOGIRI_STR_NEW2(c_entity_desc->name);
   rb_constructor_args[2] = NOKOGIRI_STR_NEW2(c_entity_desc->desc);
 
-  cNokogiriHtmlEntityDescription = rb_const_get_at(mNokogiriHtml, rb_intern("EntityDescription"));
-  return rb_class_new_instance(3, rb_constructor_args, cNokogiriHtmlEntityDescription);
+  cNokogiriHtml4EntityDescription = rb_const_get_at(mNokogiriHtml4, rb_intern("EntityDescription"));
+  return rb_class_new_instance(3, rb_constructor_args, cNokogiriHtml4EntityDescription);
 }
 
 void
 noko_init_html_entity_lookup()
 {
-  cNokogiriHtmlEntityLookup = rb_define_class_under(mNokogiriHtml, "EntityLookup", rb_cObject);
+  cNokogiriHtml4EntityLookup = rb_define_class_under(mNokogiriHtml4, "EntityLookup", rb_cObject);
 
-  rb_define_method(cNokogiriHtmlEntityLookup, "get", get, 1);
+  rb_define_method(cNokogiriHtml4EntityLookup, "get", get, 1);
 }

data/ext/nokogiri/{html_sax_parser_context.c → html4_sax_parser_context.c}

@@ -1,6 +1,6 @@
 #include <nokogiri.h>
 
-VALUE cNokogiriHtmlSaxParserContext ;
+VALUE cNokogiriHtml4SaxParserContext ;
 
 static void
 deallocate(xmlParserCtxtPtr ctxt)
@@ -110,10 +110,10 @@ void
 noko_init_html_sax_parser_context()
 {
   assert(cNokogiriXmlSaxParserContext);
-  cNokogiriHtmlSaxParserContext = rb_define_class_under(mNokogiriHtmlSax, "ParserContext", cNokogiriXmlSaxParserContext);
+  cNokogiriHtml4SaxParserContext = rb_define_class_under(mNokogiriHtml4Sax, "ParserContext", cNokogiriXmlSaxParserContext);
 
-  rb_define_singleton_method(cNokogiriHtmlSaxParserContext, "memory", parse_memory, 2);
-  rb_define_singleton_method(cNokogiriHtmlSaxParserContext, "file", parse_file, 2);
+  rb_define_singleton_method(cNokogiriHtml4SaxParserContext, "memory", parse_memory, 2);
+  rb_define_singleton_method(cNokogiriHtml4SaxParserContext, "file", parse_file, 2);
 
-  rb_define_method(cNokogiriHtmlSaxParserContext, "parse_with", parse_with, 1);
+  rb_define_method(cNokogiriHtml4SaxParserContext, "parse_with", parse_with, 1);
 }

data/ext/nokogiri/{html_sax_push_parser.c → html4_sax_push_parser.c}

@@ -1,6 +1,6 @@
 #include <nokogiri.h>
 
-VALUE cNokogiriHtmlSaxPushParser;
+VALUE cNokogiriHtml4SaxPushParser;
 
 /*
  * call-seq:
@@ -88,8 +88,8 @@ void
 noko_init_html_sax_push_parser()
 {
   assert(cNokogiriXmlSaxPushParser);
-  cNokogiriHtmlSaxPushParser = rb_define_class_under(mNokogiriHtmlSax, "PushParser", cNokogiriXmlSaxPushParser);
+  cNokogiriHtml4SaxPushParser = rb_define_class_under(mNokogiriHtml4Sax, "PushParser", cNokogiriXmlSaxPushParser);
 
-  rb_define_private_method(cNokogiriHtmlSaxPushParser, "initialize_native", initialize_native, 3);
-  rb_define_private_method(cNokogiriHtmlSaxPushParser, "native_write", native_write, 2);
+  rb_define_private_method(cNokogiriHtml4SaxPushParser, "initialize_native", initialize_native, 3);
+  rb_define_private_method(cNokogiriHtml4SaxPushParser, "native_write", native_write, 2);
 }

data/ext/nokogiri/libxml2_backwards_compat.c

@@ -20,14 +20,14 @@ xmlFirstElementChild(xmlNodePtr parent)
     return (NULL);
   }
   switch (parent->type) {
-  case XML_ELEMENT_NODE:
-  case XML_ENTITY_NODE:
-  case XML_DOCUMENT_NODE:
-  case XML_HTML_DOCUMENT_NODE:
-    cur = parent->children;
-    break;
-  default:
-    return (NULL);
+    case XML_ELEMENT_NODE:
+    case XML_ENTITY_NODE:
+    case XML_DOCUMENT_NODE:
+    case XML_HTML_DOCUMENT_NODE:
+      cur = parent->children;
+      break;
+    default:
+      return (NULL);
   }
   while (cur != NULL) {
     if (cur->type == XML_ELEMENT_NODE) {
@@ -57,20 +57,20 @@ xmlNextElementSibling(xmlNodePtr node)
     return (NULL);
   }
   switch (node->type) {
-  case XML_ELEMENT_NODE:
-  case XML_TEXT_NODE:
-  case XML_CDATA_SECTION_NODE:
-  case XML_ENTITY_REF_NODE:
-  case XML_ENTITY_NODE:
-  case XML_PI_NODE:
-  case XML_COMMENT_NODE:
-  case XML_DTD_NODE:
-  case XML_XINCLUDE_START:
-  case XML_XINCLUDE_END:
-    node = node->next;
-    break;
-  default:
-    return (NULL);
+    case XML_ELEMENT_NODE:
+    case XML_TEXT_NODE:
+    case XML_CDATA_SECTION_NODE:
+    case XML_ENTITY_REF_NODE:
+    case XML_ENTITY_NODE:
+    case XML_PI_NODE:
+    case XML_COMMENT_NODE:
+    case XML_DTD_NODE:
+    case XML_XINCLUDE_START:
+    case XML_XINCLUDE_END:
+      node = node->next;
+      break;
+    default:
+      return (NULL);
   }
   while (node != NULL) {
     if (node->type == XML_ELEMENT_NODE) {
@@ -101,14 +101,14 @@ xmlLastElementChild(xmlNodePtr parent)
     return (NULL);
   }
   switch (parent->type) {
-  case XML_ELEMENT_NODE:
-  case XML_ENTITY_NODE:
-  case XML_DOCUMENT_NODE:
-  case XML_HTML_DOCUMENT_NODE:
-    cur = parent->last;
-    break;
-  default:
-    return (NULL);
+    case XML_ELEMENT_NODE:
+    case XML_ENTITY_NODE:
+    case XML_DOCUMENT_NODE:
+    case XML_HTML_DOCUMENT_NODE:
+      cur = parent->last;
+      break;
+    default:
+      return (NULL);
   }
   while (cur != NULL) {
     if (cur->type == XML_ELEMENT_NODE) {

data/ext/nokogiri/nokogiri.c

@@ -1,8 +1,10 @@
 #include <nokogiri.h>
 
 VALUE mNokogiri ;
-VALUE mNokogiriHtml ;
-VALUE mNokogiriHtmlSax ;
+VALUE mNokogiriGumbo ;
+VALUE mNokogiriHtml4 ;
+VALUE mNokogiriHtml4Sax ;
+VALUE mNokogiriHtml5 ;
 VALUE mNokogiriXml ;
 VALUE mNokogiriXmlSax ;
 VALUE mNokogiriXmlXpath ;
@@ -13,38 +15,39 @@ VALUE cNokogiriXmlCharacterData;
 VALUE cNokogiriXmlElement;
 VALUE cNokogiriXmlXpathSyntaxError;
 
-void noko_init_xml_attr();
-void noko_init_xml_attribute_decl();
-void noko_init_xml_cdata();
-void noko_init_xml_comment();
-void noko_init_xml_document();
-void noko_init_xml_document_fragment();
-void noko_init_xml_dtd();
-void noko_init_xml_element_content();
-void noko_init_xml_element_decl();
-void noko_init_xml_encoding_handler();
-void noko_init_xml_entity_decl();
-void noko_init_xml_entity_reference();
-void noko_init_xml_namespace();
-void noko_init_xml_node();
-void noko_init_xml_node_set();
-void noko_init_xml_processing_instruction();
-void noko_init_xml_reader();
-void noko_init_xml_relax_ng();
-void noko_init_xml_sax_parser();
-void noko_init_xml_sax_parser_context();
-void noko_init_xml_sax_push_parser();
-void noko_init_xml_schema();
-void noko_init_xml_syntax_error();
-void noko_init_xml_text();
-void noko_init_xml_xpath_context();
-void noko_init_xslt_stylesheet();
-void noko_init_html_document();
-void noko_init_html_element_description();
-void noko_init_html_entity_lookup();
-void noko_init_html_sax_parser_context();
-void noko_init_html_sax_push_parser();
-void noko_init_test_global_handlers();
+void noko_init_xml_attr(void);
+void noko_init_xml_attribute_decl(void);
+void noko_init_xml_cdata(void);
+void noko_init_xml_comment(void);
+void noko_init_xml_document(void);
+void noko_init_xml_document_fragment(void);
+void noko_init_xml_dtd(void);
+void noko_init_xml_element_content(void);
+void noko_init_xml_element_decl(void);
+void noko_init_xml_encoding_handler(void);
+void noko_init_xml_entity_decl(void);
+void noko_init_xml_entity_reference(void);
+void noko_init_xml_namespace(void);
+void noko_init_xml_node(void);
+void noko_init_xml_node_set(void);
+void noko_init_xml_processing_instruction(void);
+void noko_init_xml_reader(void);
+void noko_init_xml_relax_ng(void);
+void noko_init_xml_sax_parser(void);
+void noko_init_xml_sax_parser_context(void);
+void noko_init_xml_sax_push_parser(void);
+void noko_init_xml_schema(void);
+void noko_init_xml_syntax_error(void);
+void noko_init_xml_text(void);
+void noko_init_xml_xpath_context(void);
+void noko_init_xslt_stylesheet(void);
+void noko_init_html_document(void);
+void noko_init_html_element_description(void);
+void noko_init_html_entity_lookup(void);
+void noko_init_html_sax_parser_context(void);
+void noko_init_html_sax_push_parser(void);
+void noko_init_gumbo(void);
+void noko_init_test_global_handlers(void);
 
 static ID id_read, id_write;
 
@@ -152,12 +155,14 @@ void
 Init_nokogiri()
 {
   mNokogiri         = rb_define_module("Nokogiri");
+  mNokogiriGumbo    = rb_define_module_under(mNokogiri, "Gumbo");
+  mNokogiriHtml4     = rb_define_module_under(mNokogiri, "HTML4");
+  mNokogiriHtml4Sax  = rb_define_module_under(mNokogiriHtml4, "SAX");
+  mNokogiriHtml5    = rb_define_module_under(mNokogiri, "HTML5");
   mNokogiriXml      = rb_define_module_under(mNokogiri, "XML");
-  mNokogiriHtml     = rb_define_module_under(mNokogiri, "HTML");
-  mNokogiriXslt     = rb_define_module_under(mNokogiri, "XSLT");
-  mNokogiriXmlXpath = rb_define_module_under(mNokogiriXml, "XPath");
   mNokogiriXmlSax   = rb_define_module_under(mNokogiriXml, "SAX");
-  mNokogiriHtmlSax  = rb_define_module_under(mNokogiriHtml, "SAX");
+  mNokogiriXmlXpath = rb_define_module_under(mNokogiriXml, "XPath");
+  mNokogiriXslt     = rb_define_module_under(mNokogiri, "XSLT");
 
   rb_const_set(mNokogiri, rb_intern("LIBXML_COMPILED_VERSION"), NOKOGIRI_STR_NEW2(LIBXML_DOTTED_VERSION));
   rb_const_set(mNokogiri, rb_intern("LIBXML_LOADED_VERSION"), NOKOGIRI_STR_NEW2(xmlParserVersion));
@@ -213,6 +218,13 @@ Init_nokogiri()
 #endif
 
   xmlInitParser();
+  exsltRegisterAll();
+
+  if (xsltExtModuleFunctionLookup((xmlChar*)"date-time", EXSLT_DATE_NAMESPACE)) {
+    rb_const_set(mNokogiri, rb_intern("LIBXSLT_DATETIME_ENABLED"), Qtrue);
+  } else {
+    rb_const_set(mNokogiri, rb_intern("LIBXSLT_DATETIME_ENABLED"), Qfalse);
+  }
 
   cNokogiriSyntaxError = rb_define_class_under(mNokogiri, "SyntaxError", rb_eStandardError);
   noko_init_xml_syntax_error();
@@ -257,6 +269,7 @@ Init_nokogiri()
   noko_init_xml_document_fragment();
   noko_init_xml_document();
   noko_init_html_document();
+  noko_init_gumbo();
 
   noko_init_test_global_handlers();
 

data/ext/nokogiri/nokogiri.h

@@ -1,7 +1,7 @@
 #ifndef NOKOGIRI_NATIVE
 #define NOKOGIRI_NATIVE
 
-#if _MSC_VER
+#ifdef _MSC_VER
 #  ifndef WIN32_LEAN_AND_MEAN
 #    define WIN32_LEAN_AND_MEAN
 #  endif /* WIN32_LEAN_AND_MEAN */
@@ -15,7 +15,7 @@
 #  include <windows.h>
 #endif
 
-#if _WIN32
+#ifdef _WIN32
 #  define NOKOPUBFUN __declspec(dllexport)
 #  define NOKOPUBVAR __declspec(dllexport) extern
 #else
@@ -69,6 +69,7 @@ xmlNodePtr xmlLastElementChild(xmlNodePtr parent);
 #include <ruby/st.h>
 #include <ruby/encoding.h>
 #include <ruby/util.h>
+#include <ruby/version.h>
 
 #define NOKOGIRI_STR_NEW2(str) NOKOGIRI_STR_NEW(str, strlen((const char *)(str)))
 #define NOKOGIRI_STR_NEW(str, len) rb_external_str_new_with_enc((const char *)(str), (long)(len), rb_utf8_encoding())
@@ -92,12 +93,16 @@ xmlNodePtr xmlLastElementChild(xmlNodePtr parent);
 
 
 NOKOPUBVAR VALUE mNokogiri ;
-NOKOPUBVAR VALUE mNokogiriHtml ;
-NOKOPUBVAR VALUE mNokogiriHtmlSax ;
+NOKOPUBVAR VALUE mNokogiriGumbo ;
+NOKOPUBVAR VALUE mNokogiriHtml4 ;
+NOKOPUBVAR VALUE mNokogiriHtml4Sax ;
+NOKOPUBVAR VALUE mNokogiriHtml5 ;
 NOKOPUBVAR VALUE mNokogiriXml ;
 NOKOPUBVAR VALUE mNokogiriXmlSax ;
+NOKOPUBVAR VALUE mNokogiriXmlXpath ;
 NOKOPUBVAR VALUE mNokogiriXslt ;
 
+NOKOPUBVAR VALUE cNokogiriEncodingHandler;
 NOKOPUBVAR VALUE cNokogiriSyntaxError;
 NOKOPUBVAR VALUE cNokogiriXmlAttr;
 NOKOPUBVAR VALUE cNokogiriXmlAttributeDecl;
@@ -128,10 +133,11 @@ NOKOPUBVAR VALUE cNokogiriXmlXpathContext;
 NOKOPUBVAR VALUE cNokogiriXmlXpathSyntaxError;
 NOKOPUBVAR VALUE cNokogiriXsltStylesheet ;
 
-NOKOPUBVAR VALUE cNokogiriHtmlDocument ;
-NOKOPUBVAR VALUE cNokogiriHtmlSaxPushParser ;
-NOKOPUBVAR VALUE cNokogiriHtmlElementDescription ;
-NOKOPUBVAR VALUE cNokogiriHtmlSaxParserContext;
+NOKOPUBVAR VALUE cNokogiriHtml4Document ;
+NOKOPUBVAR VALUE cNokogiriHtml4SaxPushParser ;
+NOKOPUBVAR VALUE cNokogiriHtml4ElementDescription ;
+NOKOPUBVAR VALUE cNokogiriHtml4SaxParserContext;
+NOKOPUBVAR VALUE cNokogiriHtml5Document ;
 
 typedef struct _nokogiriTuple {
   VALUE         doc;
@@ -177,7 +183,8 @@ VALUE noko_xml_node_set_wrap(xmlNodeSetPtr node_set, VALUE document) ;
 
 VALUE noko_xml_document_wrap_with_init_args(VALUE klass, xmlDocPtr doc, int argc, VALUE *argv);
 VALUE noko_xml_document_wrap(VALUE klass, xmlDocPtr doc);
-NOKOPUBFUN VALUE Nokogiri_wrap_xml_document(VALUE klass, xmlDocPtr doc); /* deprecated. use noko_xml_document_wrap() instead. */
+NOKOPUBFUN VALUE Nokogiri_wrap_xml_document(VALUE klass,
+    xmlDocPtr doc); /* deprecated. use noko_xml_document_wrap() instead. */
 
 #define DOC_RUBY_OBJECT_TEST(x) ((nokogiriTuplePtr)(x->_private))
 #define DOC_RUBY_OBJECT(x) (((nokogiriTuplePtr)(x->_private))->doc)